NISTA wins GDS hackathon with AI project-planning tool Cub

TL;DR:

  • The National Infrastructure and Service Transformation Authority (NISTA) has won the Government Digital Service (GDS) AI Engineering Hackathon with Cub, a tool that turns project initiation documents (PIDs) and early scoping material into a fully formed project plan.
  • Cub generates an interactive Kanban board, resource and cost tracking, a live risk register, a chatbot and document viewer, and supplier intelligence with sentiment analysis from news on similar live projects. It is underpinned by the Teal Book of UK government project best practice.
  • Resultsense view: this is exactly the bottom-up, in-pathway AI tooling UK government has been talking about scaling for two years — built in a single day at a hackathon, by a Department for Transport-adjacent infrastructure body. UK SMEs in govtech should treat Cub as the template the Cabinet Office is most likely to want replicated.

NISTA lead data scientist Jess Grindlay framed the brief succinctly: “projects don’t go wrong, they start wrong”. Cub is positioned as the early-pathway intervention that helps a project team set up correctly from day one.

What Cub does

The tool ingests a project’s initiation document and any early scoping material, then produces:

  • A full project plan structured against the Teal Book (UK government’s project-management standard)
  • An interactive Kanban board with resource and cost tracking
  • A live risk register that updates as the project plan changes
  • A chatbot and document viewer to interrogate both the PID and the generated plan
  • Advanced analytics, including supplier intelligence and sentiment analysis drawn from news on similar live projects

The supplier-intelligence and sentiment layer is the more unusual piece — most AI project-planning tools stop at structure and don’t reach into ongoing market signal on suppliers and comparable projects.

How the hackathon worked

The GDS AI Engineering Hackathon, hosted by Version 1 at CodeNode in London, brought together 40 teams comprising 200 data scientists and engineers from across UK government. Teams were given access to AI tooling and engineering support and judged on innovation, collaboration, use of AI, testing, edge-case handling and plans for future development.

The format — single-day, multi-team, full-stack prototypes — mirrors the GDS approach to digital service development: rapid prototyping over long-cycle procurement. That GDS now runs the format for AI tools specifically signals where the agency is taking the discipline.

Why this matters for UK govtech

NISTA is the result of merging the Infrastructure and Projects Authority with elements of the National Infrastructure Commission. It oversees major UK government infrastructure delivery — projects where early-stage planning failures historically translate into the worst public-sector cost overruns. A Cub-style tool, if generalised, sits squarely in the path of those failures.

The interesting commercial question is how (and whether) Cub becomes available to UK SMEs and the wider public sector. The Teal Book is publicly available; the AI tooling stack inside Cub will likely use commodity components plus government-specific data. UK govtech vendors should expect a procurement signal in the next two quarters as the Cabinet Office considers whether to scale Cub-like tools across departments.

UK relevance

Two takeaways for UK readers. First, in-pathway AI tooling — built quickly inside government, using existing standards — is starting to outpace bespoke external procurement in measurable ways. Second, suppliers that map their products to UK government playbooks (Teal Book, Service Standard, Technology Code of Practice) will find the path to public-sector deployment shorter than vendors entering with US-defined product categories.

Looking forward

The watchpoint is whether NISTA publishes Cub publicly or as an internal-only government tool. GDS’s open-source tradition argues for the former; NHS England’s same-week move to lock down public GitHub repos shows the political weather has shifted. Either decision will tell UK SMEs and govtech vendors which way Whitehall’s AI-tooling posture is heading in 2026.

Share this article